This proposal addresses the problem of coding speech and audio signals for transmission, while preserving the integrity of background information that can be useful to intelligence analysts. In Phase I, the feasibility of a wavelet approach l was demonstrated by comparison to other, commonly used methods. In Phase II, the algorithms will be tuned to particular application requirements, and a real time device will be developed and tested.A variety of speech coding algorithms at several different transmission rates are currently available. However, most of these approaches are based on specific models of speech production. Background interference, which generally does not match the production model incorporated in the algorithm, can be severely degraded, and yield little information to listeners. The automatic identification of background noises, (e.g. identifying an aircraft from the nature of the engine noise in the background of a speech transmission) is an area of increased interest in the C3I community. This activity requires maintenance of a high degree of integrity of background sounds during speech compression.
Keywords: C3i Background Identification Wavelets Speech Coding
